增强特征融合的动态图卷积的机载LiDAR点云分类

针对动态图卷积神经网络(dynamic graph convolutional neural network, DGCNN)聚合邻居点信息时的局限性，提出一种增强特征融合的动态图卷积神经网络模型EFF-DGCNN，并应用于机载LiDAR点云分类。该模型主要基于DGCNN提出特征增强模块和特征融合模块，对原始三维点云进行分类。首先，基于DGCNN对原始点云进行边缘卷积获取局部特征和全局特征；然后，将全局特征集成于各层的局部特征得到增强局部特征，据此凸显点云不同特征的重要性，使网络更加关注有利于分类的特征；最后，对不同增强局部特征进行特征融合得到深层次特征，从而实现点云的分类。为验证所提模型的分类性能，在GML_DataSetA数据集和ISPRS数据集分别进行了点云分类试验。试验结果表明：相比于DGCNN，所提EFF-DGCNN模型具有更好的分类能力，能更好地区分结构相似的点云。

Abstract

Aiming at the limitations of dynamic graph convolutional neural network (DGCNN) on aggregating neighbor point information, a dynamic graph convolutional neural network based on enhanced feature fusion (EFF-DGCNN) model is proposed and is used for airborne LiDAR point cloud classification. The model presents the feature enhancement module and the feature fusion module based on DGCNN, which can be applied to the classification of original 3D point clouds. Firstly, the local and global features of the original point clouds are obtained by edge convolution based on DGCNN. Then, the global features are integrated into the local features of each layer to enhance the local features, so as to highlight the importance of different features of point clouds and make the network pay more attention to the features conducive to classification. Finally, different enhanced local features are fused to obtain deep features. The fused enhanced local features are used for classification of airborne LiDAR point clouds. In order to verify the classification performance of the proposed model, experiments are conducted on the GML_DataSetA dataset and ISPRS dataset. It is demonstrated that compared with DGCNN, the proposed EFF-DGCNN model has better classification ability and can better distinguish point clouds with similar structures.

参考文献

[1] 张利明. 机载激光雷达点云数据分类方法研究［D］. 成都: 西南交通大学, 2013.ZHANG L M. Research on classification method of airborne lidar point cloud data［D］.Chengdu: Southwest Jiaotong University, 2013.

[2] JAKOVLJEVIC G, GOVEDARICA M, ALVAREZ-TABOADA F, et al. Accuracy assessment of deep learning based classification of LiDAR and UAV points clouds for DTM creation and flood risk mapping［J］. Geosciences, 2019, 9(7): 323.

[3] 朱庆, 李世明, 胡翰, 等. 面向三维城市建模的多点云数据融合方法综述［J］. 武汉大学学报(信息科学版), 2018, 43(12): 1962-1971.ZHU Q, LI S M, HU H, et al. Multiple point clouds data fusion method for 3D city modeling［J］. Geomatics and Information Science of Wuhan University, 2018, 43(12): 1962-1971.

[4] 秦和娟, 管海燕, 李迪龙. 利用张量投票的机载LiDAR点云道路骨架线提取［J］. 计算机工程与应用, 2020, 56(18): 193-201.QIN H J, GUAN H Y, LI D L. Road centerlines extraction of airborne LiDAR point clouds data based on tensor voting algorithm［J］. Computer Engineering and Applications, 2020, 56(18): 193-201.

[5] 曾远, 陈亦, 姚攀, 等. 基于机载激光点云数据的输电线走廊自动识别技术［J］. 应用激光, 2021, 41(5): 1033-1038.ZENG Y, CHEN Y, YAO P, et al. Automatic identification technology of transmission line corridor based on airborne laser point cloud data［J］. Applied Laser, 2021, 41(5): 1033-1038.

[6] 王丹菂, 徐青, 邢帅, 等. 一种由粗到精的机载激光测深信号检测方法［J］. 测绘学报, 2018, 47(8): 1148-1159.WANG D D, XU Q, XING S, et al. A detection method of airborne laser sounding signal from coarse to fine［J］. Acta Geodaetica et Cartographica Sinica, 2018, 47(8): 1148-1159.

[7] 赵传, 郭海涛, 卢俊, 等. 基于深度残差网络的机载LiDAR点云分类［J］. 测绘学报, 2020, 49(2): 202-213.ZHAO C, GUO H T, LU J, et al. Airborne LiDAR point cloud classification based on deep residual network［J］. Acta Geodaetica et Cartographica Sinica, 2020, 49(2): 202-213.

[8] 潘锁艳, 管海燕. 机载多光谱LiDAR数据的地物分类方法［J］. 测绘学报, 2018, 47(2): 198-207.PAN S Y, GUAN H Y. Object classification using airborne multispectral LiDAR data［J］. Acta Geodaetica et Cartographica Sinica, 2018, 47(2): 198-207.

[9] 孙杰, 赖祖龙. 利用随机森林的城区机载LiDAR数据特征选择与分类［J］. 武汉大学学报(信息科学版), 2014, 39(11): 1310-1313.SUN J, LAI Z L. Airborne LiDAR feature selection for urban classification using random forests［J］. Geomatics and Information Science of Wuhan University, 2014, 39(11): 1310-1313.

[10] 朱江涛, 黄睿. 基于 Adaboost 的高光谱与 LiDAR 数据特征选择与分类［J］. 遥感信息, 2014, 29(6): 68-72.ZHU J T, HUANG R. Feature selection and classification of hyperspectral data and LiDAR data based on adaboost［J］. Remote Sensing Information, 2014, 29(6): 68-72.

[11] 胡迎香, 高红旗, 夏万求, 等. 机载雷达点云亚热带针叶林单木分割探究［J］. 应用激光, 2021, 41(6): 1301-1309.HU Y X, GAO H Q, XIA W Q, et al. Study on single tree segmentation of subtropical coniferous forest with airborne radar point cloud［J］. Applied Laser, 2021, 41(6): 1301-1309.

[12] 杨俊涛, 康志忠. 多尺度特征和马尔可夫随机场模型的电力线场景点云分类法［J］. 测绘学报, 2018, 47(2): 188-197.YANG J T, KANG Z Z. Multi-scale features and Markov random field model for powerline scene classification［J］. Acta Geodaetica et Cartographica Sinica, 2018, 47(2): 188-197.

[13] HU X Y, YUAN Y. Deep-learning-based classification for DTM extraction from ALS point cloud［J］. Remote Sensing, 2016, 8(9): 730.

[14] RIZALDY A, PERSELLO C, GEVAERT C, et al. Ground and multi-class classification of airborne laser scanner point clouds using fully convolutional networks［J］. Remote Sensing, 2018, 10(11): 1723.

[15] ZHAO R B, PANG M Y, WANG J D. Classifying airborne LiDAR point clouds via deep features learned by a multi-scale convolutional neural network［J］. International Journal of Geographical Information Science, 2018, 32(5): 960-979.

[16] TE G S, HU W, ZHENG A M, et al. RGCNN: Regularized graph CNN for point cloud segmentation［C］//Proceedings of the 26th ACM international conference on Multimedia. Seoul, Republic of Korea. New York: ACM, 2018: 746-754.

[17] QI C R, SU H, MO K C, et al. PointNet: deep learning on point sets for 3D classification and segmentation［C］//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, HI, USA: IEEE, 2017: 77-85.

[18] QI C R, YI L, SU H, et al. PointNet++: Deep hierarchical feature learning on point sets in a metric space［C］// Advances in Neural Information Processing Systems(NIPS 2017). New York: Curran Associates, Inc, 2017: 5105-5114.

[19] 赵中阳, 程英蕾, 释小松, 等. 基于多尺度特征和PointNet的LiDAR点云地物分类方法［J］. 激光与光电子学进展, 2019, 56(5): 052804.ZHAO Z Y, CHENG Y L, SHI X S, et al. Terrain classification of LiDAR point cloud based on multi-scale features and PointNet［J］. Laser & Optoelectronics Progress, 2019, 56(5): 052804.

[20] 释小松, 程英蕾, 赵中阳. 利用神经网络的城区机载激光雷达点云分类算法［J］. 计算机应用研究, 2020, 37(4): 1256-1260.SHI X S, CHENG Y L, ZHAO Z Y. Airborne LiDAR point clouds classification algorithm for urban classification using neural network［J］. Application Research of Computers, 2020, 37(4): 1256-1260.

[21] CHEN Y, LIU G L, XU Y M, et al. PointNet++ network architecture with individual point level and global features on centroid for ALS point cloud classification［J］. Remote Sensing, 2021, 13(3): 472.

[22] WANG Y, SUN Y B, LIU Z W, et al. Dynamic graph CNN for learning on point clouds［J］. ACM Transactions on Graphics, 2019, 38(5): 1-12.

[23] SHAPOVALOV R, VELIZHEY A, BARINOVA O. Non-associative markov networks for 3d point cloud classification［J］. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2010, 38(A): 103-108.

[24] WU W X, QI Z A, LI F X. PointConv: deep convolutional networks on 3D point clouds［C］//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach, CA, USA: IEEE, 2020: 9613-9622.

[25] WEN C C, YANG L N, LI X, et al. Directionally constrained fully convolutional neural network for airborne LiDAR point cloud classification［J］. ISPRS Journal of Photogrammetry and Remote Sensing, 2020, 162: 50-62.

余锦, 刘智慧, 方琮淇, 赖祖龙. 增强特征融合的动态图卷积的机载LiDAR点云分类[J]. 应用激光, 2023, 43(6): 0132. Yu Jin, Liu Zhihui, Fang Congqi, Lai Zulong. Airborne LiDAR Point Cloud Classification Based on Dynamic Graph Convolutionwith Enhanced Feature Fusion[J]. APPLIED LASER, 2023, 43(6): 0132.

增强特征融合的动态图卷积的机载LiDAR点云分类

关于本站 Cookie 的使用提示

全站搜索

增强特征融合的动态图卷积的机载LiDAR点云分类

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索